Research On Fault Diagnosis Method For Doubly-fed Generator Of Wind Turbine Based On BSA-PF

The doubly-fed generator is an important component of the wind turbine generator system,and its stable and healthy operation has an extremely important influence on the whole unit.However,when the doubly-fed is installed in the mountains,desert or the sea,the operating environment will become worse,which results in frequent failures of wind turbines in various forms,and short-circuit between stator windings is a common fault in double-fed generator.At the same time,electricity,voltage and other sensors will fail,which may cause the unit to stop,and it may also cause damage to the relevant electrical facilities.If the failure is known in time,we can arrange relevant maintenance works can to recover the unnecessary economic loss.Therefore,it is of great significance to study the fault diagnosis of doubly-fed generators.The fault diagnosis method based on particle filter algorithm is a kind of model-based fault diagnosis method,which is very suitable for dealing with the fault diagnosis problem of nonlinear system.Therefore,the optimization of particle filter algorithm is also a hot topic for scholars at home and abroad.To solve the existing problems of the particle filter algorithm,this paper applied the improved particle filter to the fault diagnosis of doubly-fed generator of wind turbine.The main works are as follow:(1)The doubly-fed generator system of the wind turbine has a complicated structure,and the rotor speed of doubly-fed generator is non-linear,which leads doubly-fed generator system has the characteristics of nonlinear and strong coupling.Thus,this paper analyzes the working principle,failure mechanism and feature of the doubly-fed generator system structure deeply,based on the mathematical equations of doubly-fed generator system,to study the variation trend of parameters of stator windings,current and voltage sensors under different fault conditions,confirm the fault type and fault characteristic parameters reasonably.Finally,we establish the multifault diagnosis model based on different parameters.(2)As the the standard particle filter algorithm has low fault diagnosis accuracy due to sample dilution,we adopt a new algorithm to optimize the particle filter based on improved bird swarm algorithm.Firstly,we introduce the dynamic adaptive coefficient and the adaptive step to solve the problem that the standard bird swarm algorithm is easy to fall into the local optimum,the proposed algorithm applies the position and global optimal position information of each bird to the adaptive change control,so as to improve the problem of falling into local optimal.Secondly,we use the improved bird swarm algorithm to optimize the particle filter resampling process,which makes the particles move to the high likelihood region by simulating the foraging,warning and flight behavior of the birds,so that it can improve the particle depletion and improve the accuracy of the system state estimation.(3)In this paper,we proposed a fault isolation method combining Fuzzy C-Means(FCM)clustering algorithm with multiple models to solve the problem that the fault isolation strategy is single and cannot achieve effective isolation.Firstly,based on the multi-fault model of doubly-fed generator system,we apply the improved bird swarm optimization particle filter algorithm to the state estimation of doubly-fed generators,so that we can obtain the residual by comparing the state estimation value of the improved particle filter with the actual output value of the system,and the conduct fault detection.Finally,we calculate the membership degree of the the actual system output and the clustering center of each fault model,which can judge the fault types and improve the accuracy of fault separation.